# dynamic-feed MCP server

94 keyless tools of live, verifiable data for AI: weather, hazards, space, CVEs. Ed25519-signed.

## Links
- Registry page: https://www.getdrio.com/mcp/ai-dynamicfeed-dynamic-feed
- Website: https://dynamicfeed.ai

## Install
- Endpoint: https://dynamicfeed.ai/mcp
- Auth: Not captured

## Setup notes
- Remote endpoint: https://dynamicfeed.ai/mcp
- Remote endpoint: https://dynamicfeed.ai/sse

## Tools
- search_ai_models (Search AI Models) - Search current AI models by price, context window, and capability.

Use this for up-to-date model pricing/features you don't reliably know.
Prices are USD per 1M tokens. Results are cheapest-input-price first.
Args:
  query: match part of a model name/id (e.g. "haiku", "gpt").
  provider: filter to one provider (openai, anthropic, google, xai, mistral, deepseek, groq).
  max_input_price: only models at or below this USD/1M input price.
  min_context: only models with at least this context window (tokens).
  needs_vision: only models that accept images.
  limit: max results.
Envelope: this searches our model-pricing registry, so measured_at = when the freshest matching
row was last refreshed (each row's `updated_at`); max_age 18h covers the 12h registry-refresh
cycle so a current row never falsely reads "stale". A search returning nothing yields
unavailable — there's no honest observation time to claim.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- airport_delays (Airport Delays) - Live FAA airport delays, ground stops, closures & ground-delay programs (US NAS, keyless). `airport`
= 3-letter code (LGA, LAX…) or blank for all current disruptions. Ground stops change minute-to-minute —
pure post-cutoff state. An empty/short list means the airspace is running normally.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- drug_shortages (Drug Shortages) - US drug shortages from the FDA (openFDA, keyless). `query` = drug/generic name (amoxicillin…) or blank
for the most recently updated. Is drug X in shortage right now — status/reason/company. A model's stale
yes/no here is clinically misleading; confirm clinically.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- sanctions_screen (Sanctions Screen) - Screen a name against the US OFAC SDN + UK Sanctions List (FCDO, Open Government Licence) — keyless.
`name` = party to check; optional `program` (an OFAC program e.g. IRAN/CUBA, or a UK regime/origin).
Each result is tagged with its `list` (US-OFAC | UK). A match means the party is sanctioned — legally
binding the instant it's posted, so models can't know recent designations. Confirm the full entry first.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- company_lookup (Company Lookup) - Resolve a company name (or exact 20-char LEI) to its authoritative global Legal Entity Identifier +
canonical legal name, address, status and local registration, from GLEIF (CC0 public domain, keyless).
The deterministic join key to sanctions_screen and sec_filings — an LLM can't reliably map a name to a
real, currently-active registered entity. Pass {"query":"Apple Inc"} or {"lei":"HWUPKR0MPOU8FGXBT394"};
optional `country` (ISO-2) filter. Confirm against the official register before acting.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- ephemeris (Ephemeris) - Solar & lunar timing for a place + day — sunrise/sunset, the full twilight set (civil/nautical/
astronomical), moonrise/moonset, moon phase + illumination — from the U.S. Naval Observatory (public
domain, keyless). Pass a `location` (city, geocoded) OR `lat`/`lon`; optional `date` (YYYY-MM-DD, default
today) and `tz` offset (default 0=UTC). For agent/robot daylight & moonlight gating, golden-hour windows
and night-ops planning — an LLM has no clock or location.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- emissions (Emissions) - National greenhouse-gas emissions for a country (ISO-3 code, e.g. USA/CHN/IND) + year, from Climate
TRACE (CC BY 4.0, keyless): CO2/CH4/N2O + CO2-equivalent (100yr & 20yr GWP), the country's global rank
and its share of world emissions. Climate TRACE's own modelled national inventory; annual, lags ~1-2yr.
For climate/ESG/policy agents that need a citeable emissions figure an LLM can't reliably recall.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- federal_register (Federal Register) - Recent US Federal Register documents — rules, proposed rules, notices, executive orders (keyless).
`doc_type` = RULE | PRORULE | NOTICE | PRESDOCU. Answers 'what regulation took effect / what did the
President sign' — total post-cutoff blindness for models.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- drought (Drought) - US drought severity — % of area in each D0-D4 class by week (U.S. Drought Monitor, keyless).
`area` = 'total' (US national) or a state FIPS code. D0=abnormally dry … D4=exceptional drought. Weekly.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- water_levels (Water Levels) - Real-time US river levels — gage height (ft) + streamflow/discharge (ft3/s) from USGS NWIS (keyless).
US only — see au_water for the Australian (BoM Water Data Online) counterpart.
Use for "how high / how fast is river X right now", flood risk (rising gage) or drought (low flow).
Provide a 2-letter US `state`, USGS `site` number(s) comma-separated, or neither for major rivers.
Args:
  state: 2-letter US state code (e.g. TX, CA).
  site: USGS site number(s), comma-separated.
  limit: max sites.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- sports_pulse (Sports Pulse) - Global ATTENTION + official schedule for a sporting event, team or competition — e.g. the 2026 FIFA
World Cup. Returns the event's hosts/start-end dates/sport plus a worldwide attention signal: daily
Wikipedia article views by language edition, with 7-day momentum, peak and a per-language breakdown.
Use for "how much buzz is event X getting / where in the world / is interest rising". This is the
NEUTRAL attention layer (Wikimedia Pageviews + Wikidata, CC0) — NOT live scores, fixtures or odds.
Args:
  topic: event/team/competition, resolved via Wikidata (default '2026 FIFA World Cup').
  days: attention window, 7-90 (default 30).
  lang: primary Wikipedia language edition (en, es, pt, fr, de, ...).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- world_cup (World Cup) - 2026 FIFA World Cup FIXTURES + final RESULTS — all 104 matches, the next `upcoming` fixtures and the
`recent` most-recent final scores, plus progress (played / remaining). Source: openfootball (PUBLIC
DOMAIN, redistributable). Results appear post-match as the open dataset updates — NOT live minute-by-
minute scores (those are licensed) and NO betting odds. Pair with sports_pulse for global attention.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- product_recalls (Product Recalls) - US product recalls — food, drug & medical-device enforcement reports (openFDA, keyless, CC0).
`category` = all | food | drug | device; `query` matches the product description. Most-recent first;
Class I = most serious. Use for "is product/brand X recalled right now". Informational — confirm
against the official FDA recall notice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- treasury_yields (Treasury Yields) - US Treasury daily par yield curve (1M-30Y) + the 10Y-2Y / 10Y-3M spreads — the classic recession
signal (an inverted curve has preceded recessions). Today's curve is post-cutoff; a model can't know it.
Use for "what's the yield curve / is it inverted / current 10-year Treasury". Keyless, US public domain.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- weather_alerts (Weather Alerts) - Active US NWS weather alerts/warnings — tornado, flood, severe thunderstorm, heat, winter,
air-quality, marine. `area` = 2-letter US state (CA, TX, FL...) or marine zone; blank = nationwide
Severe/Extreme. `event` optionally filters by type. Life-safety data that expires in hours; keyless,
US public domain. Pairs with current_weather/weather_forecast.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- carbon_intensity (Carbon Intensity) - Great Britain grid carbon intensity right now (gCO2/kWh + index) + the live fuel mix + the
cleanest upcoming half-hour from the 24h forecast. Use to time energy-heavy work / answer "is the
grid clean now?". Keyless, CC-BY. GB only — see grid_carbon for Australia (per-NEM-state) + GB.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- sec_filings (Sec Filings) - Recent SEC EDGAR filings for a US public company by `ticker` (AAPL, NVDA, TSLA...). `form`
optionally filters: 8-K (material event), 10-K/10-Q (financials), S-1 (IPO), 4 (insider trade).
Filed continuously — a model can't know the recent ones. Keyless, US public domain.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- ai_models (AI Models) - Current & trending AI MODELS from the open-model ecosystem (Hugging Face) — name, org, task,
popularity (likes/downloads) and release date. Use for "what AI models are trending / newest / what's
the latest <X> model". This is the OPEN side (Llama, Qwen, DeepSeek, Mistral, Gemma, Phi…); for the
closed flagships (GPT, Claude, Gemini, Grok) with pricing & versions use search_ai_models.
Args:
  query: search a model name (e.g. llama, qwen, whisper).
  org: filter by org/author (e.g. meta-llama, deepseek-ai, Qwen, mistralai, google).
  task: text-generation (default), text-to-image, automatic-speech-recognition, … or 'any'.
  sort: trending (default) | newest | downloads.
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- service_status (Service Status) - Live up/down/degraded status for major AI & dev services (OpenAI, Anthropic,
GitHub, Cloudflare, etc.). Use this to answer "is X up right now?". Services
with issues are listed first.
Args:
  category: filter by ai | dev | infra | platform.
  only_issues: only return services currently degraded or down.
  limit: max results.
Envelope: measured_at = when the freshest status row was last probed (each row's `updated_at`);
max_age 300s matches the ~5-min probe cadence, so freshness is honest live/stale.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- ai_provider_latency (AI Provider Latency) - Measured latency & uptime for each major AI provider's API (OpenAI, Anthropic,
Google, Mistral, Groq, DeepSeek, xAI) — Dynamic Feed's own measurements, fastest
first. Use this to pick the fastest/most-reliable provider right now, or to route.
Args:
  window_hours: rolling window for the averages (default 24).
Envelope: this is OUR measurement probe, so measured_at = when we last probed (the freshest
row's `last_checked`); max_age 15 min covers the ~10-min probe cadence, so a current reading
is "live". No samples in the window yields unavailable.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- software_version (Software Version) - Latest stable versions & end-of-life (EOL) dates for software — languages, runtimes,
frameworks, databases, and OSes (Python, Node.js, PostgreSQL, React, Kubernetes, Ubuntu, ...).
Use this instead of guessing: "what's the latest stable version of X?" and "is version Y still
supported or end-of-life?". Each result has `latest_version`, `eol_date`, and a `status`
(supported / nearing-eol / end-of-life), plus a provenance envelope. Source: endoflife.date.
Args:
  product: exact product slug (e.g. python, nodejs, postgresql, react, kubernetes).
  query: match a product name if you don't know the exact slug.
  category: language | runtime | framework | database | os | tool.
  only_supported: exclude end-of-life release lines.
  only_lts: only long-term-support lines.
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- security_advisories (Security Advisories) - Recent software security advisories / CVEs — each with the affected package, vulnerable
version range, the patched version that fixes it, severity, and CVSS score. Use this to check
if a package has a recent advisory, or to get the latest critical CVEs. Pairs with
software_version (is my stack current AND safe?). Newest first. Source: GitHub Advisory Database.
Note: covers recently-published reviewed advisories, not the full historical CVE corpus.
Envelope: this is an EVENT feed, so checked_at = when WE last refreshed the advisory store
(freshness reflects how current our mirror is, NOT how long since the last CVE — a quiet stretch
is not stale data). The newest advisory's own age is surfaced as latest_advisory_age_s.
Args:
  query: match summary / package / CVE id / GHSA id.
  package: affected package name (e.g. lodash, requests, log4j).
  ecosystem: npm | pip | maven | go | rubygems | nuget | composer | rust | ...
  severity: low | moderate | high | critical.
  min_cvss: minimum CVSS score (0-10).
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- market_hours (Market Hours) - Returns current open/closed status and next open/close times for major exchanges, crypto (24/7), and forex (24/5). Appropriate when a user asks whether a trading venue or market session is open.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- interest_rates (Interest Rates) - Live US benchmark interest rates — Effective Fed Funds Rate, SOFR, OBFR, etc. from the
New York Fed. Use instead of guessing the current Fed/benchmark rate (training data is stale).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- satellites (Satellites) - Satellite/constellation catalog with each satellite's altitude, orbital period, and a derived
speed-of-light round-trip latency to ground (the physics floor for orbital networks / data
centers in space). Use for "how many Starlink satellites / orbital latency".
Args:
  group: starlink, stations, gps-ops, oneweb, galileo, weather, active, geo, science.
  limit: max satellites in the sample.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- satellite_position (Satellite Position) - Real-time position of ANY satellite by its NORAD id OR a common name — computed with SGP4 from the
latest live TLE: exact latitude/longitude, altitude, velocity, ground-footprint radius, the
speed-of-light latency floor, and how old the TLE is. Use for "where is the ISS / Hubble right now"
or to get a precise sub-satellite point. This is the TRUE propagated position, not an approximation.
Args:
  norad_id: NORAD catalog number (e.g. 25544 = ISS, 20580 = Hubble, 44714 = a Starlink).
  name: a common name instead of the id — "ISS", "Hubble", "Tiangong". Pass one of norad_id or name.
  track_minutes: also return a predicted ground-track this many minutes ahead (0 = just the current
    position; max 360).
  step_seconds: ground-track sampling step in seconds (default 60).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- rocket_launches (Rocket Launches) - Upcoming or recently-launched orbital rocket launches from Launch Library 2 (The Space
Devs). Per launch: name, provider (+ type/country), vehicle (+ family/reusability/flight
count), pad + location (+ coordinates), net launch time + a live countdown (T-minus / T-plus),
net precision + launch window, mission (name/type/orbit/description), and current go/scrub
status. `window` = "upcoming" (default — next launches by net time) | "previous" (recently
launched). `limit` = how many to return, 1-30 (default 10). Keyless; Launch Library 2 free
to use (commercial OK) — we re-serve a value-added envelope (countdown, normalised fields,
freshness + provenance), never the raw passthrough or the non-commercial launch imagery.
Schedules slip frequently; trust `status` + `net_precision`.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- gov_contracts (Gov Contracts) - US federal contract & assistance awards from USAspending.gov (keyless, US public domain). Who was
awarded what, by which agency, for how much, over what period, the authoritative public spending
record routinely cross-examined in audits, FOIA and litigation, so a decision about a vendor or
programme can be backed by a signed, contemporaneous record. args (all optional): agency (awarding
top-tier agency name contains), recipient (vendor name contains), keyword (free-text on description),
award_type "contract" (default) | "grant" | "all", min_amount (USD floor), limit 1-50. Re-served as a
signed, provenance-stamped envelope; we witness the public record, we never move or award funds.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- spacex (Spacex) - Live SpaceX activity — next & recent SpaceX launches + Starlink constellation size, from public
data; not financial/stock data. A single composite digest answering "what is SpaceX doing right
now": the next SpaceX launches and the most recent ones (name, vehicle, pad, net launch time +
live countdown, go/scrub status), plus the size of the Starlink constellation we track (object
count + a couple of sample objects). Argless. INDEPENDENT — not affiliated with or endorsed by
SpaceX; "SpaceX"/"Starlink" used descriptively only. Reuses two feeds we already serve (no new
upstream): rocket_launches (Launch Library 2, The Space Devs) filtered to the SpaceX provider, and
satellites(group=starlink) off our CelesTrak TLE mirror — so it does NOT hammer upstreams on the
request path. Each part degrades independently (one source down → the others still return).
HARD CONSTRAINT: NO stock price, share price, valuation, or IPO/investment data of any kind —
launch schedule + constellation facts only, informational, from public data (The Space Devs,
CelesTrak).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- orbital_data_centers (Orbital Data Centers) - DATA CENTERS IN SPACE — curated registry of compute/AI spacecraft in orbit (Starcloud's NVIDIA
H100 GPU, ESA Φsat-2 AI edge, D-Orbit in-orbit cloud), each enriched with LIVE orbital data
(altitude, period, inclination) and the speed-of-light round-trip latency floor for ground links.
Use for "what data centers / compute are in space, and the latency to reach them". Unique data.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- space_weather (Space Weather) - Full space-weather now-cast (NOAA SWPC) — the NOAA R/S/G storm scales (radio blackout /
radiation storm / geomagnetic storm), geomagnetic Kp index, solar wind (speed, density, and Bz —
the storm driver), the latest solar X-ray flux / flare, and an aurora-visibility hint. Affects
satellite comms & drag, GPS accuracy, HF radio, power grids, and the radiation/power environment
for orbital compute. Use for "is there a geomagnetic/solar storm right now".

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- internet_outages (Internet Outages) - Live BGP routing health for a network resource — an ASN (e.g. "AS3215"), an IP ("8.8.8.8"), or a
prefix ("193.0.0.0/22") — from RIPEstat (RIPE NCC's open routing-information service). Returns global
visibility (how many of RIPE's route collectors currently see the resource) + an outage signal:
healthy ≥0.9 · degraded ≥0.5 · outage <0.5. A sharp visibility drop = the network is losing global
reachability. Use for "is network/ASN X reachable right now?". Pass `resource`.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- gps_interference (Gps Interference) - Probable GPS/GNSS interference (jamming/spoofing) over an area, MEASURED from live ADS-B navigation
integrity (the GPSJam method) — unique to Dynamic Feed, derived from adsb.lol (CC0). Pass a region
preset (london · blacksea · baltic · easternmed · persiangulf · korea · finland · caucasus · northsea)
OR lat+lon (+ radius_km). A high degraded_fraction = many aircraft losing GPS-grade navigation =
probable interference (none <0.1 · low ≥0.1 · medium ≥0.25 · high ≥0.5).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- energy_grid (Energy Grid) - Live US electricity grid data by region/operator (US EIA, public domain). region = US48 (national)
or a grid operator: CAISO · PJM · MISO · ERCOT · ISONE · NYISO · SPP · BPA. metric = demand |
generation | mix (current fuel mix in MW). Hourly. Pairs with carbon_intensity.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- global_macro (Global Macro) - Worldwide macro indicators for ANY country (World Bank Open Data, CC-BY). country = an ISO code
(US, GB, DE, IN, BR, JP, CN, ZA…) or name. indicator = gdp · gdp_growth · gdp_per_capita ·
cpi_inflation · unemployment · population · exports · imports · current_account · fx_reserves · fdi ·
govt_debt_pct_gdp · all. Globalizes the China-only macro to the whole world.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- earthquakes (Earthquakes) - Recent earthquakes worldwide from USGS — magnitude, depth, location, PAGER impact alert (green→
red, distinct from raw magnitude), tsunami flag, felt reports and significance. Use for "recent /
biggest earthquakes" or "was there a quake near X". Sorted by impact then magnitude.
Args:
  min_magnitude: minimum magnitude (default 2.5).
  period: hour | day | week | month (default day).
  limit: max results.
  tsunami_only: only quakes flagged with tsunami potential.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- natural_events (Natural Events) - Active natural events worldwide from NASA EONET — wildfires, severe storms (hurricanes/typhoons),
volcanoes, floods, sea/lake ice. Use for "what natural disasters/events are happening right now".
Args:
  category: wildfires | severeStorms | volcanoes | floods | seaLakeIce (blank = all).
  days: look-back window in days (default 30).
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- wildfires (Wildfires) - Active US wildfires from NIFC WFIGS — incident name, size in acres, % contained, cause, state and
location. The 'how big / how contained' detail satellite hotspot feeds can't give (pairs with
natural_events, which has global fire points but not size/containment). Largest first.
Args:
  state: optional US state filter, e.g. CA, TX (2-letter).
  only_uncontained: only fires below 100% containment.
  min_acres: minimum incident size in acres.
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- global_disasters (Global Disasters) - Active worldwide disaster alerts (GDACS) — ONE cross-hazard feed across earthquakes, tropical
cyclones, floods, droughts and volcanoes, each with a Green/Orange/Red humanitarian-impact level
and the affected countries. Use for "what major disasters are happening in the world right now / in
country X". The global severity layer complementing earthquakes, wildfires and natural_events.
Args:
  alert_level: levels to include — "Orange;Red" (default, impactful) or "Green;Orange;Red" (all).
  event_type: EQ | TC | FL | DR | VO | WF (blank = all).
  country: country name or ISO3 code (e.g. PHL, Philippines).
  days: look-back window in days (default 7).
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- volcanoes (Volcanoes) - Current US volcano alert levels (USGS Volcano Hazards Program, keyless, public domain). `status`:
'elevated' (default — only volcanoes at YELLOW/ORANGE/RED or ADVISORY/WATCH/WARNING) or 'all' (every
USGS-monitored volcano). Each result carries the ground `alert_level`, aviation `color_code`, the
observatory, when the notice was issued and the official notice URL, most-severe first. SCOPE: US
volcanoes only — the global Smithsonian GVP dataset is licensed non-commercial, so it's not included.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- tides (Tides) - Current water level + next high/low tides for a US coastal station (NOAA CO-OPS, keyless, public
domain). Pass a NOAA `station` id (e.g. 9414290) or a `location` preset (san francisco, new york, boston,
seattle, honolulu, miami, galveston, …); `units` english|metric. Heights are vs the MLLW datum. SCOPE:
US stations only. Not for navigation — confirm against official NOAA tide tables.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- exploited_vulnerabilities (Exploited Vulnerabilities) - Vulnerabilities CONFIRMED exploited in the wild — the CISA Known Exploited Vulnerabilities (KEV)
catalog, each enriched with its EPSS exploitation-probability score (FIRST.org), plus a risk
`summary`. The 'which CVEs actually matter right now' layer that pairs with security_advisories
(disclosed CVEs) and check_vulnerability (is package@version affected).
Args:
  query: match a CVE id / name / description.
  vendor: filter by vendor — accepts a COMMA LIST (e.g. "Microsoft, Cisco, Fortinet"), match ANY.
  product: filter by product — accepts a COMMA LIST (e.g. "Windows, PAN-OS"), match ANY.
  stack: a client's whole tech stack as a comma list (e.g. "Fortinet, Ivanti, Citrix, Apache"),
    matched against vendor OR product — returns exactly the actively-exploited CVEs affecting that
    stack: a live per-stack vulnerability watch (see the `summary` for the headline risk).
  ransomware_only: only CVEs linked to known ransomware campaigns.
  recent_days: only CVEs added to the KEV catalog in the last N days.
  sort: "recent" (newest catalogued first, default) or "risk" (highest EPSS first).
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- live_flights (Live Flights) - Live aircraft positions right now — real-time flight tracking from open ADS-B (adsb.lol, CC0).
Answers "what's flying over X?" and "where is flight Y?". Airborne flights first.
Args:
  region: a preset area — london, newyork, losangeles, paris, tokyo, dubai, singapore,
          sydney, frankfurt, hongkong, morocco, spain, europe, usa, asia.
  callsign: filter by flight callsign / number (e.g. BAW123, UAL456).
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- current_weather (Current Weather) - Current weather anywhere — by city name or lat/lon. Returns temperature, feels-like, wind,
humidity, precipitation and conditions, live. Use instead of guessing the weather.
Args:
  city: city name (e.g. London, Tokyo, New York) — geocoded automatically.
  lat: latitude (alternative to city).
  lon: longitude (alternative to city).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- weather_forecast (Weather Forecast) - Multi-day weather forecast (up to 16 days) for a city or lat/lon — daily high/low, precipitation,
wind, conditions. Use for "what's the weather this week / will it rain".
Args:
  city: city name (geocoded). lat/lon: alternative. days: 1-16 (default 7).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- air_quality (Air Quality) - Current air quality for a city or lat/lon — PM2.5, PM10, US & European AQI, ozone, NO2,
with a category (Good/Moderate/Unhealthy...). Use for "is the air quality safe in X".
Single-network (model) — see air_quality_fusion for a two-network cross-validated reading.
Args:
  city: city name (geocoded). lat/lon: alternative.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- check_vulnerability (Check Vulnerability) - Is this exact package (and version) vulnerable? FULL historical lookup across the entire
OSV.dev corpus (Google) — every matching advisory + the versions that fix it. Use to check a
dependency: "does lodash 4.17.10 have any known CVEs?".
Args:
  package: package name (lodash, requests...). For Maven, pass the FULL 'group:artifact'
           coordinate (e.g. org.apache.logging.log4j:log4j-core) — a bare artifact name matches
           nothing in OSV, so it returns a hint, never a misleading "not vulnerable".
  version: exact version (optional but recommended, e.g. 4.17.10).
  ecosystem: npm | pip | maven | go | rubygems | nuget | cargo | composer (optional).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- china_data (China Data) - Returns current World Bank macro series for China, including GDP, growth, CPI, trade, M2, unemployment, and FDI, with source and observation metadata. Appropriate for current macroeconomic context.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- shipping_status (Shipping Status) - Global shipping & maritime data AIs lack. CURRENTLY LIVE: real-time Baltic vessel positions (AIS)
enriched with ship name, type and destination (scope=vessels, the default). The 28 maritime-chokepoint
transits (Suez, Panama, Hormuz, Malacca, Taiwan...) and the global port registry from IMF PortWatch are
COMING SOON (a future release) and return a `note` until then. Use for "what ships are near Z".
Args:
  scope: vessels (default, live) | chokepoints | ports.
  query: filter by chokepoint/port name or country, or ship name/destination (vessels).
  region: Baltic preset for scope=vessels (baltic, helsinki, stockholm, kiel, gdansk, copenhagen, riga, gulf-of-finland).
  limit: max results.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- whats_changed_since_your_training (Whats Changed Since Your Training) - Returns source-backed developments about a user-specified topic that occurred after a model's training period or knowledge cutoff. Appropriate when the user explicitly asks about recent, current, or post-cutoff developments.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- reality_check (Reality Check) - Fact-check a statement against LIVE data — the anti-hallucination tool. Pass any claim about the
current world ("the latest Python is 3.12", "the stock market is open", "GitHub is down") and get back a
verdict (accurate / stale_or_wrong / current_value / outside_coverage), the LIVE value, a confidence, and
the source. Compound claims (joined by "and") are split and each part checked. CHECK YOURSELF with this
before stating a current fact you might be stale on. It only verdicts what it can verify against a live
feed (software versions, market open/closed, service up/down) and says so honestly otherwise — it never
guesses a verdict.
Args:
  claim: the statement to verify, in plain language.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- detect_drift (Detect Drift) - DRIFT DETECTION — audit a batch of AI-stated facts against live ground-truth in one call. Pass
`claims` (a list of statements an LLM produced) or `text` (a paragraph that's split into claims); each
is fact-checked and graded — drifted / confirmed / unverifiable, with a severity — plus a summary
(drifted_count, overall). Use it to self-audit an answer before sending it.
Args:
  claims: list of statements to check.
  text: alternatively, a paragraph that gets split into claims.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- current_time (Current Time) - The current date & time anywhere — an LLM has no internal clock, so use this instead of guessing
"today" or "now". tz is an IANA timezone (e.g. "America/New_York", "Asia/Tokyo"); default UTC. Returns
UTC + local time, date, weekday, weekend flag, ISO week and unix time.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- convert (Convert) - Convert an amount between currencies at the LIVE rate (not a stale training-era rate). CURRENTLY
LIVE: fiat (USD/EUR/GBP/JPY/CNY/AUD/...). Crypto (BTC/ETH/SOL/...) and precious metals (XAU/XAG/XPT/XPD)
are coming in a future release. e.g. 100 USD→JPY, 5000 EUR→GBP.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- countdown (Countdown) - Time until a LIVE event you can't compute without a clock: a market open/close (NYSE, NASDAQ, LSE,
TSE, HKEX, SSE, ASX, XETRA) or the next rocket launch (optionally a provider, e.g. "next SpaceX
launch"). Returns a human countdown + seconds + the target time.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- world_pulse (World Pulse) - One call: what's happening on Earth RIGHT NOW — the biggest recent earthquake, the newest
actively-exploited CVE, the fastest AI API (we measure it), the trending new AI model and the latest
Python release — with a plain-English summary. The live state of the world at a glance.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- cite (Cite) - Get a fresh, CITEABLE source + timestamp for a current datapoint — so you can cite it, not guess.
Pass ANY tool, source, or topic (earthquakes, current_weather, USGS, Open-Meteo, …) for its
authoritative source + licence + attribution + verify URL, or a software product (python, nodejs, …)
for its live latest-version citation.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- package_health (Package Health) - Health & security posture of a software package (npm / PyPI / Go / Maven / Cargo / NuGet /
RubyGems) from deps.dev (Google Open Source Insights, keyless): latest version, license, count of
known security advisories, the OpenSSF Scorecard (0-10 security-posture score for the source repo +
its weakest checks) and popularity (stars/forks). The "should I depend on this?" check — pairs with
check_vulnerability (is a version vulnerable) and software_version (is the runtime current).
Args: package (e.g. "lodash", "requests"), ecosystem (npm|pypi|go|maven|cargo|nuget|rubygems),
version (optional — defaults to the latest).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- github_releases (Github Releases) - Latest releases for a GitHub repo (owner/name) — version, date, author, link — via the public
releases atom feed (keyless, no rate limit). "What's the newest release of X" — a model can't know
releases published after its training cutoff. e.g. repo="facebook/react".

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- hacker_news (Hacker News) - Live Hacker News stories (keyless) — the tech/AI attention layer right now. kind = top | new |
best; optional case-insensitive title filter (e.g. query="AI"). Returns title, score, comments, url.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- dns_lookup (Dns Lookup) - Live DNS records for a domain via Cloudflare DNS-over-HTTPS (keyless): A / AAAA / MX / TXT / NS /
CNAME / SOA / CAA. Resolve a domain to what's actually live in DNS now (not a stale cached guess).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- cert_check (Cert Check) - Live TLS/SSL certificate check for any domain — measured by a direct handshake RIGHT NOW: issuer,
validity window, days to expiry, expiring-soon flag, SANs, negotiated TLS version. Is this site's
cert valid / when does it expire / who issued it — facts, not a cached scan.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- wikipedia_trends (Wikipedia Trends) - What is the world paying ATTENTION to right now — top-viewed Wikipedia articles (the global
zeitgeist, ~1 day lag), or pass a topic for that article's 14-day attention trend with a
week-over-week change. Pageview data is CC0 (public domain) from the Wikimedia Pageviews API.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- econ_calendar (Econ Calendar) - What is SCHEDULED to move markets — upcoming US Treasury auctions (type, term, size, date) and
FOMC meeting dates incl. the next meeting. Pure schedule facts (never prices), from US Treasury
Fiscal Data + the Federal Reserve calendar (both public domain). 'When is the next Fed meeting /
next 10-year auction?'

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- climate_enso (Climate Enso) - Current EL NIÑO / LA NIÑA state — latest ONI anomaly + phase, official episode check, recent
seasons, and NOAA CPC's alert-system status (e.g. 'El Niño Watch'). The Pacific state that drives
global drought/flood/bushfire risk — direct input for climate-risk and parametric posture.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- sst_anomaly (Sst Anomaly) - Sea-surface temperature + anomaly vs climatology for any ocean point or named region
(great_barrier_reef [default], moreton_bay, tasmania, ningaloo, coral_sea, north_west_shelf,
spencer_gulf, florida_keys, gulf_of_mexico, north_sea, mediterranean) — with an honest
marine-heatwave heuristic flag (anomaly ≥ +1.0 °C). NOAA OISST v2.1 (public domain), ~1-2 day lag.
Marine heatwaves drive aquaculture losses, bleaching and cyclone energy.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- sea_ice (Sea Ice) - Latest daily Arctic + Antarctic sea-ice extent (million km²) from the NSIDC Sea Ice Index
(G02135, v4): current extent + date per pole, change vs the prior week, and the same calendar
day ~10 years earlier. region = "arctic" | "antarctic" | "both" (default "both"). Extent = ocean
with ≥15% ice concentration; near-real-time values are provisional, ~1-2 day lag. The headline
polar-climate indicator behind Arctic shipping seasons and climate-risk posture. Public domain
(NOAA@NSIDC), citation carried in provenance.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- world_events (World Events) - Latest global news-event coverage from the GDELT Project, which monitors worldwide news
in 100+ languages: the most recent articles matching `query` (keyword/phrase, optional —
matched against full article text) from outlets in `region` (country name, optional) within
the past `hours` (1-168, default 24). Argless = a global pulse of the latest coverage.
Article METADATA only — titles, links, outlets, timestamps, source countries — never bodies.
Open GDELT terms (unlimited commercial use; citation carried in provenance).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- grid_carbon (Grid Carbon) - Live grid carbon intensity (gCO2/kWh): current value, trend vs ~3 hours ago, and the live
generation mix by fuel. `region` = "au" (Australian NEM, default) | "nsw" | "qld" | "vic" |
"sa" | "tas" (single NEM state) | "uk" (Great Britain) | "all" (AU + UK). AU values are
DERIVED and honestly flagged so (method in response): live AEMO 5-minute dispatch SCADA
weighted by AEMO's own CDEII per-generator CO2e emission factors — AEMO's official daily
CDEII index rides alongside for cross-checking. GB values are NESO-measured. Complements
carbon_intensity (UK-only, cleanest-window forecast). Keyless; AEMO general permission —
any purpose with attribution — + NESO CC BY 4.0.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- au_economy (Au Economy) - Official AUSTRALIAN economic indicators from the Australian Bureau of Statistics Data API
(SDMX) — STATISTICS ONLY, NO prices. `indicator` = "cpi" (annual inflation, default) |
"cpi_quarterly" | "unemployment" | "wages" | "gdp". Returns the latest value (a %), its real
ABS reference period (e.g. 2026-Q1 / 2026-04), the unit/description, and a short trend (latest
vs prior). CPI annual inflation is DERIVED by us from the All-groups CPI index (ABS publishes
the index, not the annual rate, as its series) — flagged `derived: true` with the index
alongside for checking; unemployment, wage growth (WPI) and GDP growth are ABS's own published
rates. Source: Australian Bureau of Statistics (data.api.abs.gov.au), CC BY 4.0; keyless,
~12 h cache (ABS releases monthly/quarterly). HARD CONSTRAINT: every value is an official
statistic/index/rate (CPI index, inflation %, unemployment %, wage-growth %, GDP-growth %) —
NEVER a market price, quote, FX rate, or asset valuation. Not financial advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- us_economy (US Economy) - Official US economic indicators from the U.S. Bureau of Labor Statistics Public Data API —
STATISTICS ONLY, NO prices. `indicator` = "cpi" (annual inflation, default) | "unemployment" |
"payrolls" (monthly non-farm jobs change) | "ppi" (producer-price inflation) | "wages"
(average-hourly-earnings growth). Returns the latest value, its real BLS reference period (e.g.
2026-05), the unit/description, and a short trend (latest vs prior). CPI, PPI and wage growth are
DERIVED by us as a 12-month change from the BLS index/$ level (BLS publishes the index, not the
annual rate, as the API series) — flagged `derived: true` with the index/level alongside for
checking; unemployment is BLS's own published rate; payroll change is a month-on-month difference.
Source: U.S. Bureau of Labor Statistics (api.bls.gov), US public domain; keyless, ~12 h cache (BLS
releases monthly). HARD CONSTRAINT: every value is an official statistic/index/rate (CPI index,
inflation %, unemployment %, payroll head-count, PPI index, average hourly wage in $) — NEVER a
market price, quote, FX rate, or asset valuation. Not financial advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- scholarly_search (Scholarly Search) - Fact-check the published scholarly record via OpenAlex — METADATA ONLY, NO prices.
`query` = free text/topic ("CRISPR gene editing") OR a DOI / doi.org URL for an exact-work
lookup. `sort` = "relevance" (default) | "citations" | "newest" | "oldest". `limit` = max
works (default 10, cap 25). Each work returns its citation count, open-access status + URL,
publication venue + type, year/date, work type, lead authors (up to 5) + total author count
+ lead institution, references count, retraction flag, and the work's own OpenAlex
record-update time. Spans 240M+ works across every publisher (journals, conferences, books,
datasets, preprints), so it answers "does this paper exist / how cited is it / is it open
access / where was it published / who wrote it" — which a training snapshot gets stale or
hallucinates. DISTINCT from research_papers (arXiv preprints only, newest-first abstracts):
different source (OpenAlex), cross-publisher, citation/OA/venue metadata — no abstracts, no
full text. Source: OpenAlex (api.openalex.org, OurResearch), data is CC0 1.0 public domain;
keyless (polite pool), ~6 h cache. HARD CONSTRAINT: every value is bibliometric metadata (a
citation count, year, OA status, venue, author) — NEVER a market price, quote, or valuation.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- fda_recalls (Fda Recalls) - Recent FDA-regulated RECALLS & enforcement actions via openFDA — regulatory facts, NO prices.
`category` = "drug" (default) | "food" | "device". `query` = optional product / firm / reason
search term (e.g. "Pfizer", "peanut", "contamination"; blank = the most recent across that
category). `limit` = max items (default 15, cap 25). Per item: product description, recalling
firm, reason for recall, classification (Class I/II/III severity), status (Ongoing/Completed/
Terminated), recall-initiation date, distribution pattern, state/country, recall number. Plus
a short summary, counts by classification, total_matches, and the newest recall date surfaced
separately as `latest_recall_date`. Answers "what FDA-regulated medicines / food / medical
devices have been recalled lately, by whom, why, and how serious" — a live public record a
training snapshot can't have. DISTINCT from product_recalls (CPSC consumer goods — toys,
appliances) and drug_shortages (FDA supply gaps, not defect withdrawals). Source: openFDA
(api.fda.gov, U.S. FDA), U.S. Government work — public domain (CC0 1.0); keyless, ~1 h cache.
INFORMATIONAL ONLY — openFDA data is NOT for clinical or safety decision-making and may be
incomplete; confirm against the official FDA record before acting. HARD CONSTRAINT: every
value is a regulatory/safety fact (firm, reason, class, status, date) — NEVER a market price.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- air_quality_fusion (Air Quality Fusion) - Cross-validated air quality — TWO independent networks reading the same sky: the Open-Meteo
CAMS model (the existing air_quality source) vs real ground monitors via OpenAQ v3 (government
reference stations), with a per-pollutant (PM2.5, PM10, O3, NO2) agreement verdict
(high/moderate/low/divergent), both values + the delta, and station distance/freshness caveats —
the two numbers are never blended. `location` = an AU preset ("brisbane" default, "sydney",
"melbourne", "perth", "adelaide"), any city name (geocoded), or a "lat,lon" string; explicit
`lat`/`lon` override. Complements air_quality (single-network AQI). If one network is down the
result degrades honestly to single-source (agreement: "unverified"). Stations whose licence
disallows commercial use are auto-excluded via OpenAQ's machine-readable licence flags;
per-station provider attribution + licence carried in provenance.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- au_water (Au Water) - Australian river heights + gauge rainfall near any AU location — live CC-BY-licensed gauge
observations from the Bureau of Meteorology's Water Data Online (keyless KiWIS API); the
Australian counterpart of water_levels (US USGS). Per station: latest river level (m) with a
~3 h rising/falling/steady trend, or rainfall totals (mm over 1/6/24 h ending at the latest
reading), plus real observation timestamps + ages, distance, the data owner and the exact
attribution string. `location` = an AU preset ("brisbane" default, "sydney", "melbourne",
"adelaide", "hobart", "darwin", "canberra", "perth") or a "lat,lon" string; explicit
`lat`/`lon` override. `kind` = "river" | "rain" | "both". Only stations whose data owner is
CC-BY-listed (Table 1) in the BoM rain/river licensing statement are served — per-station
owner attribution carried in provenance, exclusion counts reported, WA coverage honestly
partial. Observational data ONLY — not a flood warning product; official flood
warnings/watches are issued by the Bureau at bom.gov.au/australia/warnings.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- flood_forecast (Flood Forecast) - 5-day river-discharge FORECAST near an Australian location — Copernicus CEMS Global Flood
Awareness System (GloFAS, operational LISFLOOD control run), retrieved daily from the CEMS
Early Warning Data Store and served from cache (the answer is instant — never waits on the
CDS queue). The forward-looking sibling of au_water: au_water = what AU rivers are DOING
(live gauges), flood_forecast = what the global model expects them to do over the next 5
days. Returns the strongest modelled river cell within ~90 km: 24 h-mean discharge (m3/s)
per day for ~5 days, the peak value + day, and a rising/falling/steady trend; values are the
MAX within ~25 km (0.25-deg max-pooled) cells. `location` = an AU preset ("brisbane"
default, "sydney", "melbourne", "adelaide", "hobart", "darwin", "canberra", "perth") or a
"lat,lon" string; explicit `lat`/`lon` override. Raw model guidance ONLY — no flood
thresholds or severity classifications are added, and this is NOT a flood warning; official
AU flood warnings/watches are issued by the Bureau of Meteorology at
bom.gov.au/australia/warnings. Attribution: contains modified Copernicus Emergency
Management Service information.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- aviation_weather (Aviation Weather) - Live aviation weather — decoded METAR observations + TAF forecasts for any ICAO airport
(NOAA Aviation Weather Center, public domain, keyless). Returns flight category (VFR / MVFR /
IFR / LIFR), winds, visibility, ceilings and the raw METAR/TAF report — the same data a pilot
sees in a preflight briefing, accessible by any AI agent or robot in one call. `station` = any
4-letter ICAO airport code (default YBBN = Brisbane; KJFK, EGLL, YSSY, OMDB, RJTT all served;
US stations fully covered; major international airports covered via ICAO/WMO distribution to
NOAA). `kind` = "metar" (default) | "taf" | "both". Cross-reference `gps_interference` for
GNSS outages affecting aircraft and `airport_delays` for FAA ground stops. Source: NOAA Aviation
Weather Center (aviationweather.gov) — US Government work, public domain (17 U.S.C. § 105).
CAVEAT: observational/forecast data only — not a substitute for an official preflight briefing
(FAA ATIS/AFIS or an authorized flight planning service).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- hurricanes (Hurricanes) - Active tropical cyclones from the NOAA National Hurricane Center (NHC) — tropical
depressions, tropical storms, and hurricanes. Per storm: name, classification, Saffir-Simpson
category, current position (lat/lon), max sustained winds (kt + km/h), minimum pressure,
movement (heading + speed), and a link to the latest public advisory and cone graphic.
`basin` = "all" (default) | "atlantic" (North Atlantic + Caribbean + Gulf of America) |
"epac" (Eastern + Central North Pacific). Off-season or quiet periods may return zero
storms — this is the correct honest state. Source: NOAA NHC (nhc.noaa.gov) — US government
public domain (17 U.S.C. § 105), keyless. Coverage: Atlantic + Eastern/Central Pacific ONLY
(not Australian region, Western Pacific, or Indian Ocean).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- bitcoin_network (Bitcoin Network) - Live Bitcoin NETWORK telemetry — infrastructure metrics only, NO price data.
`metric` = "overview" (default, full digest) | "fees" (recommended fee tiers sat/vB) |
"mempool" (unconfirmed tx count + vsize + pending fees) | "difficulty" (current + next
adjustment % + blocks until retarget) | "hashrate" (3-day average EH/s) | "halving"
(blocks remaining + estimated date + current/post-halving block subsidy). Overview returns
all sections in one call. Source: mempool.space public REST API (open-source, AGPL-3.0)
with Blockstream Esplora (MIT) as block-height fallback. Keyless, 60 s cache. HARD
CONSTRAINT: no BTC/USD or any fiat price — Bitcoin NETWORK mechanics only. Not financial
advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- lightning_network (Lightning Network) - Bitcoin LIGHTNING NETWORK aggregate statistics — network infrastructure metrics only,
NO price data. Returns node count, public channel count, total public channel capacity
(in BTC and satoshis), average + median channel capacity, the Tor / clearnet / unannounced
node split, and the average fee rate (ppm) — plus a human summary and the snapshot date.
Lightning is Bitcoin's layer-2 payment network; these are its routing/topology metrics.
Source: mempool.space public REST API (open-source Bitcoin explorer, AGPL-3.0). Keyless,
~10 min cache. HARD CONSTRAINT: capacity is a NETWORK metric in BTC/sats — never a fiat
valuation or market price. Not financial advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- ots_calendar_health (Ots Calendar Health) - Live health of the public OpenTimestamps (OTS) timestamp-anchoring calendar servers — the
aggregation-and-anchoring infrastructure behind tamper-evident, independently verifiable,
Bitcoin-anchored timestamps. Answers "are the public timestamp calendars up and still anchoring
right now?" before an agent trusts a fresh proof. `calendar` = "all" (default — probe every public
calendar and return an overall verdict) | "alice" | "bob" | "finney" (Eternity Wall) | "catallaxy".
Per calendar: live reachability + latency (our own probe), server version, pending commitments,
anchor transactions waiting for confirmation, the calendar's best-block height (its Bitcoin chain
view), average time between anchor transactions over the last week (anchoring cadence), and the
operational wallet balance funding anchor transactions; the overall verdict flags any calendar whose
height lags the consensus (a stalled chain view / paused anchoring). Source: Dynamic Feed's own
availability probes + each calendar's public status page (measured facts, like cert_check /
service_status). Keyless, 120 s cache. Operational balances are anchoring-funding health signals,
NOT market prices.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- vin_decode (Vin Decode) - What vehicle is this VIN, according to the US regulator? Decodes a Vehicle Identification
Number against NHTSA's Product Information Catalog Vehicle Listing (vPIC) — the catalog of how
every manufacturer assigns VINs, filed with NHTSA under 49 CFR Part 565 as a condition of selling
or importing a vehicle for US roads. `vin` = a 17-character VIN, or a partial VIN using NHTSA's
"*" wildcard (their documented example: 5UXWX7C5*BA). `model_year` = optional 4-digit year; NHTSA
recommends always sending it, and it disambiguates the 30-year VIN cycle. Returns the vehicle as
the manufacturer built it: make, manufacturer, model, series/trim, model year, body class,
vehicle type, doors/seats; powertrain (fuel type, electrification level, cylinders, displacement,
horsepower, drive type, transmission, EV battery/drive unit); assembly plant city/state/country;
the full factory safety-equipment list (ABS, ESC, TPMS, airbag positions, automatic emergency
braking, lane-keep, blind-spot, adaptive cruise, event data recorder, and more); and GVWR weight
class. NHTSA's own decode error codes and error text ride on every response, including whether
the check digit validates and whether the manufacturer is registered with NHTSA at all. Source:
NHTSA vPIC (vpic.nhtsa.dot.gov). Keyless, public domain, 24 h cache. SCOPE: a SPECIFICATION
decode of the manufacturer's VIN pattern — it describes what was built. It is NOT a registration,
title, ownership, lien, insurance, odometer, accident, salvage or theft record, and it contains
no information about any person. COVERAGE (NHTSA's own published figures, carried per response):
~99% from model year 1995, ~60-65% for 1980-1994, nothing before 1980.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- trader_positioning (Trader Positioning) - Who is actually positioned which way in a US futures market — from the regulator itself. The
CFTC Commitments of Traders (COT) report, published every Friday 3:30 p.m. ET, showing the long /
short / spreading positions held as of the prior Tuesday broken out by trader category, plus the
number of reporting traders behind each. Compiled by the CFTC from position reports that clearing
members and brokers are legally required to file — the one public, audited view of positioning.
`market` = a preset (default "gold"; also silver, copper, platinum, palladium, crude_oil,
natural_gas, gasoline, corn, wheat, soybeans, soybean_oil, soybean_meal, sugar, coffee, cocoa,
cotton, live_cattle, lean_hogs, sp500, nasdaq, russell, dow, vix, euro, yen, aussie, pound,
swiss_franc, canadian_dollar, mexican_peso, dollar_index, treasury_2y/5y/10y, treasury_bond,
fed_funds, sofr_3m) or any CFTC market-name fragment ("WHEAT-HRW", "FEEDER CATTLE", "MSCI EM
INDEX"). `report` = "legacy" (default — non-commercial speculators vs commercial hedgers vs
non-reportable small traders; covers EVERY market) | "disaggregated" (producer/merchant, swap
dealers, managed money, other reportables; physical commodities only). `weeks` = weeks of net-
positioning history, 1-26 (default 8). Per category: long/short/spreading contracts, net, the
CFTC's own week-over-week change, % of open interest, and trader counts — plus total open interest
and its weekly change. Source: the CFTC's own published Commitments of Traders files (www.cftc.gov). Keyless, public domain, 12 h cache.
HONESTY: COT is a LAGGED weekly report by design, not live — positions are as of Tuesday's close
and are published the following Friday, so every response carries the report's as-of date and its
age in days. HARD CONSTRAINT: positioning ONLY — contract counts, open interest, percentages and
trader counts. No price, quote, valuation or settlement data is returned by any path. Not
financial advice or a trading signal.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- marine_forecast (Marine Forecast) - Marine / surf forecast for a coastal point — current sea state + next-24-hour outlook.
Returns total significant wave height / period / direction, the swell component (long-period
energy driving surf), the locally-generated wind-wave, and sea-surface temperature, plus a
24-hour wave-height band. `location` = AU surf/maritime presets: "gold_coast" (default),
"byron", "bells_beach", "margaret_river", "sydney". OR pass `lat=` and `lon=` (decimal
degrees) for any coastal point worldwide. Source: Open-Meteo Marine Weather API (CC-BY 4.0;
underlying ocean-wave models DWD/MeteoFrance/ECMWF/NCEP/ERA5-Ocean). Keyless to you;
commercial-licensed server-side. FORECAST data, not a measurement — not a substitute for
an official marine forecast/warning (e.g. the BoM coastal-waters forecast) and must not be
the sole basis for any safety-of-life-at-sea decision.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- ocean_buoys (Ocean Buoys) - Live marine observations from NOAA NDBC ocean buoys — significant wave height & period,
wave direction, wind speed/direction/gust, barometric pressure (+ 3-hour tendency), air &
water temperature, dewpoint and visibility, each with the buoy's own observation timestamp +
age. `station` = an NDBC station id (e.g. "46059", "51001", "41001") -> that one buoy.
`near` = "lat,lon" (e.g. "37.7,-123.0") OR a coast preset (us west coast, us east coast,
gulf of mexico, hawaii, alaska, great lakes, pacific nw, new england, florida, california,
korea) -> the nearest reporting buoys (up to `limit`, default 5). No args -> a sensible
default open-ocean buoy. Single bulk-file fan-out: NDBC's whole-network latest_obs file is
fetched once and indexed (~1,300 stations), so every lookup is served from cache. COVERAGE:
NDBC is a NOAA/US network — no Australian buoys (run by BoM/AODN). Observational data only
— not a substitute for an official marine forecast/warning. Source: NOAA NDBC
(ndbc.noaa.gov) — US government public domain (17 U.S.C. § 105), keyless.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- research_papers (Research Papers) - Latest scholarly preprints from arXiv — newest-first — by category and/or keyword.
Returns up to 15 papers, each with: title, authors, truncated abstract, primary + all
categories, published/updated dates, arXiv id, abstract URL, PDF URL, and DOI / journal
reference when a published version exists. `category` = an arXiv taxonomy term (default
"cs.AI"). Common ones: cs.AI (AI), cs.LG (Machine Learning), cs.CL (NLP/LLMs), cs.CV
(Computer Vision), cs.RO (Robotics), cs.CR (Security), stat.ML, cs.MA (Multiagent). Any
valid arXiv category works — see arxiv.org/category_taxonomy. `query` = optional free-text
keyword/phrase, AND-combined with the category. Source: arXiv API (Cornell University) —
descriptive metadata is CC0 1.0 public domain (keyless, commercial use permitted). arXiv
is a PREPRINT server; most papers are not peer-reviewed.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- species_occurrences (Species Occurrences) - Recent species occurrence records from GBIF (Global Biodiversity Information Facility) —
individual field observations and museum specimens of where/when a species has been recorded.
Each record: species + scientific name, lat/lon, date, country, source dataset (+ GBIF
citation), recorder, and basis-of-record (human observation, preserved specimen, etc.).
`query` = a common name ("koala", "great white shark", "platypus") or scientific name
("Phascolarctos cinereus"); resolved against the GBIF backbone taxonomy. `near` = "lat,lon"
to restrict to records within ~100 km of that point (e.g. "-27.47,153.02" for Brisbane).
Keyless GBIF.org API. COMMERCIALLY CLEARED: filtered server-side to CC0 + CC BY 4.0 records
only; CC BY-NC records are excluded. OCCURRENCE data only — not a population census or range
map; coverage is sampling-biased and absence of records does not mean absence of the species.
Source: GBIF.org API v1 — CC0 1.0 + CC BY 4.0 (per record).

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- public_holidays (Public Holidays) - Public / statutory holidays for a country (and optional state/province/region) in a given
year — computed in-house, NOT scraped, so it's exact and offline-stable. Per holiday: date,
name, weekday, whether it's an observed-day shift, the country, and the subdivision (when one
was requested), plus the next upcoming holiday. `country` = ISO-3166 alpha-2 code (default
"AU"); common names accepted ("australia", "usa", "uk", "new zealand"). 100+ countries.
`year` = 4-digit year (default = current year). `subdiv` = optional subdivision code for
region-specific holidays — e.g. "NSW"/"VIC"/"QLD" for AU, "CA"/"NY" for US, "ENG"/"SCT" for
GB; national-level only if omitted (state lists are richer). Pairs with `market_hours`. Holiday
DATES are non-copyrightable facts; computed via the open-source `holidays` library (vacanza,
MIT) — commercial use + redistribution permitted, keyless. Informational only — confirm
against the official gazette for legal/payroll use.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- wikidata_facts (Wikidata Facts) - Structured factual statements about a named entity from Wikidata — the collaboratively-curated
knowledge base behind Wikipedia. Resolves `entity` to its Wikidata item, then returns the
headline facts: description, instance/class, key properties (e.g. for a country: capital,
population, area, head of government, currency, ISO codes; for a person: birth/death, occupation,
citizenship), plus cross-reference identifiers and the Wikidata QID. `entity` = a name to look
up ("Australia", "Albert Einstein", "Sydney Opera House", "Bitcoin"); default "Australia".
Reference/knowledge lookup, not a live-telemetry feed — values are as-current-as Wikidata's
community edits. Source: Wikidata (wikidata.org), MediaWiki Action API — CC0 1.0 (public
domain): commercial use + redistribution permitted, no key, no attribution required.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- health_surveillance (Health Surveillance) - US public-health surveillance signals from the CDC — the current level + week-over-week
direction of circulating illness, not individual case data. `topic` = "respiratory" (default —
emergency-department visit shares + clinical-lab test positivity for COVID-19 / influenza /
RSV) | "wastewater" (CDC NWSS viral-activity levels in community wastewater, an early-warning
leading indicator). Each metric carries its reporting week + trend. COVERAGE: United States
ONLY (CDC networks); surveillance is a population-level signal with reporting lag, not a
real-time case count or a diagnosis. Source: CDC (cdc.gov) NSSP + NWSS open datasets — US
Government public domain (17 U.S.C. § 105), keyless; carries CDC attribution + the standard
non-endorsement note. Informational only, not medical advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- clinical_trials (Clinical Trials) - Live clinical-trial registrations and status updates from ClinicalTrials.gov (US NIH/NLM
registry). Returns the most-recently-updated trials, each with: NCT id, title, status
(recruiting / completed / terminated / ...), phase, conditions, lead sponsor (+ class), study
type, enrollment, start / primary-completion / completion dates, location count + countries,
and first-posted / last-update dates. `condition` = disease/condition filter ("melanoma",
"type 2 diabetes", "alzheimer"). `query` = free-text search across the whole record
(intervention, sponsor, keyword — "crispr", "Pfizer", "semaglutide"); combine with `condition`
to narrow. Both empty (argless default) = the most-recently-updated notable trials registry-wide.
Keyless; ClinicalTrials.gov data is US Government public domain (17 U.S.C. § 105) — free for
commercial reuse, courtesy of the U.S. National Library of Medicine. Self-reported registry
data — informational only, not medical advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- uk_water (Uk Water) - Live UK river & water levels and flood warnings from the Environment Agency (England) —
real-time gauge readings near a place, each with current level, recent trend (rising / steady /
falling), the station's typical range, plus any active EA flood alerts/warnings for the area.
`location` = a UK place name ("london", "york", "oxford", "carlisle"; default "london"), OR
pass `lat`/`lon` (decimal degrees) for any point in England. COVERAGE: England only (the EA's
network — not Scotland/Wales/NI, which have separate agencies). Observational + alert data,
not a forecast, and not a substitute for the official GOV.UK flood-warning service for any
safety decision. Source: Environment Agency real-time flood-monitoring API — Open Government
Licence v3.0 (commercial use permitted with the EA's mandated attribution, carried in
provenance), keyless.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- global_indicator (Global Indicator) - A flagship development statistic from Our World in Data: the latest value for a country plus
a short multi-year trend, with full source attribution. ONE source, MANY indicators (breadth) —
CO2 per capita, population, fertility, urbanisation, GDP-per-capita (a development stat in PPP,
NOT a market price), extreme poverty, R&D spend, Human Development Index, literacy, internet
access, electricity access. Distinct from `global_macro` (World Bank): OWID adds the long-run
development + climate set. `indicator` = a slug/alias from the curated allowlist (default
"co2-emissions-per-capita"; aliases: co2, pop, gdp, hdi, literacy, internet, poverty, fertility,
urban, rd) — call indicator="list" for the full menu. `country` = ISO-3 code (AUS, USA, CHN,
GBR, IND, …); omit for the World aggregate. Source: Our World in Data (ourworldindata.org) —
OWID's processing layer is CC BY 4.0, keyless; every response carries BOTH OWID's attribution
AND each underlying producer's citation + licence. Only indicators whose underlying sources are
cleared for commercial re-serving (CC BY / CC BY IGO / CC0 / public domain) are served — a
fail-closed runtime gate refuses any non-redistributable indicator. Annual-ish statistics, not a
live-telemetry feed.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp
- pollen_forecast (Pollen Forecast) - Airborne pollen / allergen forecast — six pollen types (alder, birch, grass, mugwort, olive,
ragweed) in grains/m³, each with a plain-English severity band (low / moderate / high / very
high), for a hay-fever / asthma / rhinitis audience. Returns the current hour's reading plus the
NEXT-DAY peak per type, an overall worst-case level, and a summary naming the dominant allergens.
`location` = city name or preset; EU presets carry real data (london [default], paris, berlin,
madrid, rome, amsterdam, vienna, warsaw, athens, dublin), AU/world presets (brisbane, sydney,
melbourne, perth, new york, tokyo) are convenience but normally report `coverage` =
out_of_region. Or pass explicit `lat` + `lon`. COVERAGE IS EUROPE-ONLY: the CAMS pollen model
covers Europe (~4-day forecast in season); outside Europe — including all of Australia, the
Americas, Asia, Africa — there is NO pollen model and the response says so honestly
(`coverage` = out_of_region) instead of guessing. DISTINCT from `air_quality` /
`air_quality_fusion` (those serve POLLUTANTS — PM2.5/ozone/AQI; this serves BIOLOGICAL pollen).
Source: Open-Meteo Air Quality API / Copernicus Atmosphere Monitoring Service (CAMS) European
ensemble — CC BY 4.0 under Open-Meteo's paid commercial plan (keyless to you); attribution to
Open-Meteo + CAMS carried in provenance. Informational only, not medical advice.

Every value is returned in an Ed25519-signed, provenance-stamped envelope (source and observation time) you can verify offline against /.well-known/keys, no account required. Endpoint: https://dynamicfeed.ai/mcp

## Resources
Not captured

## Prompts
Not captured

## Metadata
- Owner: ai.dynamicfeed
- Version: 0.8.7
- Runtime: Sse, Streamable Http
- Transports: HTTP
- License: Not captured
- Language: Not captured
- Stars: Not captured
- Updated: Jul 26, 2026
- Source: https://registry.modelcontextprotocol.io